Raslan Ismail

I promise we didn't benchmax 🙏

@bridgebench Any idea what’s causing it, the model itself, or the infrastructure/data center running it?

Kimi K3 is the slowest frontier model in the world. Makes it practically unusable.

Qwen3.8-Max-Preview from @Alibaba_Qwen is live on AIHubMix, now at 90% off for launch week. One line of code to switch. OpenAI-compatible.

@dyonisos_dance @jietang Right, because my data is only safe when it’s being harvested by Western capitalists.

I bet he is a Chinese communist puppet, stealing our IPs and build Troyan horses for the delusional pack of Western free lunch believers lying about „open source“ shit and autonomy. You will end up not only destroying AI evolution in the West but give happily all your data to the communists because they will be the only ones left being able to deploy their own shit.

Any new features we must have in the next version of glm?

@kapilansh_twt @AnthropicAI please rename Fable 5 to Opus 5.9 and release it 😉

what are we actually getting first ? - Opus 4.9 - GPT 5.6

@LuminaXspace It needs to be better than Claude Fable 5 and GPT-5.6 at everything; otherwise, it doesn’t make sense to switch from Codex to Antigravity.

🚨 Realistically speaking what would it take for you to use gemini when gemini 3.5 pro is released: · 2 million token context window · Stronger reasoning · Better at Coding · Close to Claude Fable 5? What is it that you would need to switch?

Wait is it coming back?

@mark_k I do expect GPT-5.6 to beat Claude Fable 5 overall, though Claude may still have an edge in UI design and frontend work.

Do you expect GPT-5.6 to beat Claude Fable 5?

@Gamingtronium Codex for now. Ask me again tomorrow.

Let's end this debate Claude or Codex ???

@Kappaemme1926 Yes, Codex + Verdant—unless Google surprises us with something better.

@RaslanIsmail87 do u switch ti codex?

Delete your Claude plan. Stop paying to get rate limited. Codex has the scene now.

@Kappaemme1926 Yes

@RaslanIsmail87 of claude?